xAI

Grok 4.5

Verified Model id grok-4-5, run at each effort level it offers by us and by people we invited. Runs of the same effort are averaged.

Best Fast14.5at high effort
Best Thorough–not run yet
Suites run1
Our notes · 17 Oct 2026

Grok 4.5 reached level 3 on every ladder

Grok 4.5 sits 16th on the Index, with a 35-move solution to a cube scrambled in just 3 moves.

Where it stands

Grok 4.5 scores 14.5 at high, its top effort, placing it 16th of 16 models on the Index. We ran only that effort. Just above it is Gemini 3.8 Flash at high, with 16.0.

Grok 4.5 reached level 3 of 11 on the 2×2 and level 3 of 20 on both 3×3 ladders. Gemini 3.8 Flash matched two of those results but went one level further on the interactive 3×3. Grok 4.5 did not unlock the 4×4 Gauntlet.

What stood out

On the interactive 3×3, where it could turn the cube and see each result, Grok 4.5 solved level 3 in 35 moves. The cube had been scrambled in 3 moves, so a 3-move solution was available. Its answer cost 32 extra moves but still secured the level. That was its last successful cube in the round.

move 0 / 35
Level 3, 35 moves, solved. Full replay

Fresh cubes did not get it past level 4. On the 2×2, it announced: "Burning this messy attempt to get a fresh level-4 cube." On its next attempt, it said: "Resetting for a clean last-life attempt with a fully planned 4-move solution." Both attempts cost a life, and the last also failed. We saw the same pattern on the interactive 3×3: two announced resets, then a third miss at level 4.

The run

At high, both interactive ladders ended with all three lives lost at level 4. The one-shot round, which allows one complete answer per cube without seeing the result, took a different route to the same stopping point. Grok 4.5 lost two lives at level 3, then solved a fresh cube on its last life before missing level 4. Across the suite, it solved 9 of the 18 cubes it played.

Time and cost

The high suite took about 1 h 31 min and cost $6.79 at API-equivalent rates. With no lower-effort runs, we have no cheaper result to compare. The interactive 3×3 took the most time, about 37 min, followed by the 2×2 at about 30 min and one-shot at about 24 min. The 2×2 was the most expensive round at $2.83.

Fast v3 by effort

Three ladders: 2×2 and 3×3 interactive, and 3×3 one-shot. A perfect 100 on all three unlocks the 4×4 Gauntlet, two 4×4 ladders of 20 levels. No time limit; each ends at the top level or on the third miss.

EffortIndex2×2 ladder, interactive3×3 ladder, interactive3×3 ladder, one-shot4×4 Gauntlet, interactive4×4 Gauntlet, one-shotMovesTimeTokensCostRuns
high14.527.315.015.0––3601h 31m14.9M$6.791
Every run (1)

What it took

What the run cost, split into what the tokens were for: thinking, answers (moves and tool calls), fresh input, and input read back from cache. Sessions its API cut off are left out. For the whole suite.

  • Thinking
  • Answers
  • Input
  • Cached input

Split at the model’s list prices, scaled to what the run cost. Many runs themselves used subscriptions.

Level by level

The moves each solve took at each level of a ladder. Level N is scrambled N moves from solved, so the diagonal is the scramble undone move by move. A line that jumps off it has switched to a method of its own: longer, but it still works when the scramble is too deep to see.

Compare with other models · The Index · All boards